Dual Athlon MP 1U units

W Bauske wsb at paralleldata.com
Sat Jan 26 13:35:15 PST 2002


Velocet wrote:
> 
> On Sat, Jan 26, 2002 at 02:30:04AM -0600, W Bauske's all...
> > Velocet wrote:
> > >
> > > Whats the power dissipation of running dual 1.2 GHz Mp's? How about for
> > > 1.33Ghz regular athlons in non-SMP configs as comparison? (As well, how much
> > > heat comes off typical power supplies to run these systems?)
> > >
> >
> > My TigerMP XP1600 duals take about 1.7amps at 125v.
> >
> > Forgot the formula to convert to btu's. Vaguely remember a factor
> > of around 3.42. Not sure if that was for Watt's or VoltAmps. Assuming
> > a VA is approximately a Watt, 212.5 * 3.42 = 727 btu per system.
> >
> > At least with that you can calculate your AC load for a rack. Say 40
> > 1U's per rack, 29080 btu's. A ton of AC is 12000 btu's. So, 2.5 ton's
> > of AC per rack. Course, you have 40x1.7 amps going into the rack for
> > a power load of 68 Amps at 125v.
> >
> > Those that know the real numbers, please correct. A VA is really around
> > .7 - .8 watts, so these calculations are high by maybe 20%. Figure
> > the extra allows you to plug in the switches/peripherals/servers in addition
> > to the nodes.
> 
> not to mention the power supplies themselves (or was that part of the
> measurement of your 1.7A?)
> 

The 1.7 amps is at the plug-in. Everything is included.

It was measured while 100% busy on several different applications.
Note while sitting idle, the Athlon system still took the same amount of 
power so I suspect it doesn't idle down as was mentioned in another post. 

My P4's do idle down when not busy. They take from .85 - 1.3 amps depending
on speed of the processor when busy and about half the max when idling.

You can go to your local HW store and by a meter for this. The only trick
is you have to put the sensor around only one of the wires, not both or
it can't read the current. I split open a power cable (carefully) so I
could separate the wires and measure only one. Be careful not to nick the
wires inside if you try it. Don't want anyone getting shocked. You could
have an electrician make a special wire for you too if you want to be
safe. If you nick it, try again on a new power cord. They're cheap...

> 2.5 tons of A/C is required, that sounds right - that keeps the volume
> of air involved at neutral temperature (say 68F) - but how many CFM's
> of air are required to move the heat off the processors (well, heatsinks)
> fast enough to keep them comfortable (I dont know what comfortable is -
> 50-55C?) Is there a rule of thumb calc for that?
> 

That's related to the heatsink's thermal transfer efficiency. It'd be nice
if there were a simple gauge to determine CFM of a chassis but I don't
know of one. I've seen transfer rates on several HS manufacturer pages so
you might try look up your HS and see what it says. It was something like
degrees per watt of power I think. 

Bari might know more about it since he works with dense packaging.

Wes



More information about the Beowulf mailing list